Profiling of microRNA targets using activity-based protein profiling: linking enzyme activity to microRNA-185 function

Summary

MicroRNAs (miRNAs) act as cellular signal transducers through repression of protein translation. Elucidating targets using bioinformatics and traditional quantitation methods is often insufficient to uncover global miRNA function. Herein, alteration of protein function caused by microRNA-185 (miR-185), an immunometabolic microRNA, was determined using activity-based protein profiling (ABPP), transcriptomics and lipidomics. Fluorophosphonate-based ABPP of miR-185-induced changes to human liver cells revealed that exclusively metabolic serine hydrolase enzymes were regulated in activity with roles in lipid and endocannabinoid metabolism. Lipidomic analysis linked enzymatic changes to levels of cellular lipid species such as components of very-low density lipoprotein (VLDL) particles. Additionally, inhibition of one miR-185 target, MGLL, lead to decreased hepatitis C virus (HCV) levels in an infectious model. Overall, the approaches used here were able to identify key functional changes in serine hydrolases caused by miR-185 that are targetable pharmacologically, such that a small molecule inhibitor can recapitulate the microRNA phenotype.

Introduction

As an integral part of RNA interference (RNAi), microRNAs (miRNAs) act to prevent endogenous gene transcript translation and induce mRNA destabilization through RNA-RNA interactions. miRNAs influence a large number of cell processes, and have been shown to play important roles in host-pathogen interactions (Scaria et al., 2006). In the last decade, many in silico methods have been developed to predict targets of miRNAs via seed sequence alignment, including platforms such as miRanda (John et al., 2004) and TargetScan (Lewis et al., 2003). However, predicted seed sequence interactions in the 3’UTR do not always translate into repression of expression (Didiano and Hobert, 2006) and bioinformatics only allows prediction of direct targets of miRNAs. Techniques involving cross-linking and immunoprecipitation (CLIP) have helped to identify direct miRNA-mRNA interactions (Chi et al., 2009), yet such data still does not always accurately predict miRNA function. Complex interactions such as those in signaling cascades lead to changes in the proteome that are difficult to predict, and post-translational modifications substantially alter the functional proteomic landscape. Consequently, the outcome of a single miRNA’s expression cannot easily be foretold from its putative targets.

To deduce the functional effects of a specific miRNA, we used activity-based protein profiling (ABPP) as a screening method to determine the impact of miRNA-185 (miR-185) on a subset of the human hepatoma proteome. miRNA-185 has emerged as a modulator of immunometabolism in the liver. This miRNA was previously known to repress cholesterol uptake (Yang et al., 2014) and was recently shown to target a number of genes involved in lipid biosynthesis and metabolism regulation, leading to a lipid-depleted antiviral state during hepatitis C virus (HCV) and other viral infections (Singaravelu et al., 2015).

In the following study, we used ABPP (Fig. S1A) to characterize the activity of serine hydrolase enzymes during miR-185 signaling. The human proteome contains over 200 serine hydrolases, many of which have been shown to have important roles in metabolism (Merkel et al., 2002; Thomas et al., 2013) and infection (Pham, 2006). Furthermore, these enzymes have proven to be attractive pharmacological targets owing to their crucial roles in complex biological pathways, with multiple drugs targeting these enzymes already available on the market (Bachovchin and Cravatt, 2012). Results from quantitative ABPP involving multiplex stable isotope dimethyl labeling allowed several miR-185-modulated serine hydrolases to be identified. Interestingly, all the hits fell within the metabolic serine hydrolase family, and amongst these, not all were predicted targets of the miRNA. Classification of these enzymes showed enrichment of pathways such as lipid metabolism and endocannabinoid hydrolysis and pointed to modulation of a peroxisome proliferator-activated receptor (PPAR) by miR-185. Lipidomic analysis showed a decrease in cholesteryl ester and triglyceride species consistent with known roles of the small RNA and provided further insight into the structural variations within lipid classes resulting from the enzymatic changes observed with ABPP. Finally, pharmacological inhibition and siRNA-mediated knock-down of one of the targets, MGLL, significantly reduced HCV levels in an infectious cell model, further validating the role of the enzyme in miR-185’s antiviral effect. Overall, our results demonstrate the applicability of ABPP as a useful tool, in conjunction with systematic –omic analysis, as illustrated in summary figure 1A, for the investigation of non-coding RNA function. This study further demonstrates that miR185 has functional roles in lipid metabolism, endocannabinoid metabolism and very-low density lipoprotein (VLDL) synthesis.

Figure 1.

miR-185 alters protein activity and lipid metabolism A) miR-185 exerts its effects both directly, by binding to mRNAs and modifying the proteome, or indirectly by affecting downstream post-translational modifications and interactions. A combination of transcriptomics, ABPP and lipidomics analysis allows probing of the pathway at intermediate and end point states. B) Structure of fluorophosphonate (FP)-biotin. C) Workflow of the mass-spectrometry ABPP experiment. The active serine hydrolases are labelled with FP-biotin, enriched with streptavidin beads and separated from the rest of the proteome. They are then digested with trypsin before being isotopically labelled and analyzed via LC-MS/MS. D) Pie-chart depicting classification of the known serine hydrolases based on sequence relatedness (Bachovchin and Cravatt, 2012). All miR-185 regulated enzymes are metabolic. E) Average relative activity of serine hydrolases from miRNA-185 transfected cells compared to a non-targeting control RNA as determined with mass spectrometry analysis. Hits with changes between 0.8 and 1.2 were excluded (n=3). *indicates predicted miR-185 targets and §, validated miR-185 target. Data for each trial can be found in Table S1. F) Metabolic serine hydrolases with changes lower than 0.8 (red) or higher than 1.2 (blue) in activity (determined with FP-ABPP), expression (determined with microarray transcriptome analysis), or both. *One enzyme in purple, CPVL, was seen to increase in mRNA abundance and decrease in activity. †indicates enzymes that were not detected with the other method. Fold changes for the microarray data can be found in Table S2.

Results

Activity-based protein profiling reports on microRNA function

Given that miR-185 has been shown to modulate metabolism in the liver to give rise to an antiviral state, we wanted to identify enzymes which contribute to this immunometabolic effect. Of the 238 currently known serine hydrolases, about half are considered metabolic enzymes, including lipases, peptidases, esterases, and amidases (Fig. 1D) (Bachovchin and Cravatt, 2012). This large enzyme class can be characterized by ABPP using fluorophosphonate (FP) reactive groups coupled to reporter tags such as biotin (Fig. 1B) (Liu et al., 1999). In order to identify differentially regulated metabolic serine hydrolases, we transfected Huh7.5 human hepatoma cells with a miR-185 mimic and after a 72 hours incubation, lysed the cells to collect the active proteome. We incubated this lysate with a FP-biotin probe and isolated the labelled enzymes with streptavidin bead enrichment, column purification and trypsin digestion (Fig. 1C). Mass spectrometry analysis of activity-labelled serine hydrolases in the presence or absence of miR-185 in these cells yielded a total of 17 enzymes represented in three trials with average activity fold changes under 0.8 or above 1.2 (Fig. 1E, Table S1). Interestingly, only two of the hits (ACOT1 and PREPL) were strongly predicted to be direct targets of the miRNA through the miRanda and TargetScan platforms. Figure 1F compares the direction of the activity changes of serine hydrolases determined via ABPP with their transcript abundance as determined with microarray analysis. We identified six enzymes with similar changes in activity and transcript abundance, ten enzymes that were seen to only decrease in activity, and one enzyme that had lower activity while exhibiting a slight increase in mRNA abundance. These differences illustrate the ability of ABPP to report on functional changes in enzymes which may be due to post-translational modifications or interactions with other cellular factors. To obtain a visual idea of the changes in activity, we labelled miR-185 transfected lysates with fluorophosphonate attached to a fluorescent tag, FP-TAMRA. After running these samples on an SDS-PAGE gel, a change in the activity of FASN and MGLL was observable (Fig. S1B). However, due to the low abundance of some serine hydrolases as well as the poor sensitivity of this technique, other changes were not identified. Working with the hits list obtained via mass spectrometry, activity changes of enzymes with known metabolic functions and deemed of interest in the context of miR-185 were further validated with western-blot pull-downs and their change in expression was confirmed via RT-qPCR. They were then assigned into one of two categories based on their functions: lipid metabolism or endocannabinoid metabolism.

miRNA-185 downregulates the activity of serine hydrolases involved in lipid metabolism

In order to determine which of the serine hydrolases identified by ABPP might be relevant to lipid metabolism, we classified them based on their putative functions. Approximately half of the hits we found to be involved in pathways previously shown to be relevant to miRNA-185 function. Amongst the enzymes that were seen to decrease in activity, fatty acid synthase (FASN), arylacetamide deacetylase (AADAC), acyl-CoA thioesterase 1 (ACOT1) and lipase C (LIPC) are proteins well known for their roles in lipid metabolic pathways. These enzymes are involved in fatty acid biosynthesis, triglyceride hydrolysis, fatty acid β-oxidation and lipoprotein remodeling respectively (Chang and Borensztajn, 1993; Hunt et al., 2014; Lo et al., 2010; Maier et al., 2008).

To validate the changes observed via mass-spectrometry, we labelled proteomes of miR-185 transfected cells with FP-biotin, performed streptavidin pull-downs and ran the labelled fractions on an SDS-PAGE gel, followed by transfer and immunodetection for these specific enzymes (Fig. 2A). This technique provides a visual confirmation of activity changes which can be compared with abundance changes in the whole lysates. We also probed the levels of messenger RNA using RT-qPCR. We observed a reduction in both the activity and abundance of the FASN, AADAC and ACOT1 enzymes (Fig.1E, 2B, C). ACOT1 was the only one predicted to be a target of miR-185 by both the miRanda and TargetScan platforms. Interestingly, our results show a downregulation of LIPC activity by miR-185 (Fig. 1E, 2B) while the abundance of the protein was not seen to vary significantly via western blotting and had a modest decrease in mRNA (Fig. 2B, C), pointing to an enzymatic regulation mechanism which is affected by miR-185. Therefore, the reduction in the activities of these four enzymes at various steps of lipid synthesis and degradation is one of the ways in which the non-coding RNA changes the lipid profile of cells.

Figure 2.

miR-185 acts on genes involved in lipid regulation and endocannabinoid metabolism. A) Workflow of the western-blot ABPP experiments. The active serine hydrolases are labelled with FP-biotin, enriched with streptavidin beads and separated from the rest of the proteome. They are then heat denatured and run on a SDS-PAGE gel, followed by transfer and immunoblotting for the target of interest. B) Western blot analysis of FP-biotin pull-down in miR-185 transfected Huh7.5 cells. The left column shows blotted enzymes which were labelled by FP-biotin and pulled-down with streptavidin beads, representing their change in activity. The right column shows the abundance of the same proteins in their respective whole lysate input before streptavidin enrichment. SCD1 is a validated non-hydrolase miR-185 target. PTP1D was used as loading control in the input blots. C) Graph of miR-185-induced mRNA changes of hit genes ± S.D. as analyzed by RT-qPCR. n= 4, except for ABHD6, ACOT1 and LIPC, where n=5 D) PPAR alpha mRNA fold change as determined by RT-qPCR in microRNA-185 transfected cells, n=5. Significance assessed with two-tailed, unpaired student’s t-test, * = p<0.05, ** = p<0.01.

miRNA-185 regulates the activity of serine hydrolases involved in endocannabinoid lipid metabolism

To further delineate the role of miR-185, we grouped three enzymes with a common link to endocannabinoid metabolism. Functions of the endocannabinoid pathway vary based on tissue localization, but from a metabolic perspective, activation of the receptors has been linked to obesity (Sun et al., 2014), hepatic insulin signaling (Chanda et al., 2012) and hepatitis C (Patsenker et al., 2015; Sun et al., 2014). The system is comprised of the G-protein coupled receptors CB1 and CB2, their endogenous lipid ligands anandamide (N-arachidonoylethanolamine) and 2-arachidonoylglycerol (2-AG), and the enzymes which act upon them (Di Marzo et al., 2004). Breakdown of 2-AG is modulated through the activity of a few serine hydrolases, amongst them monoglyceride lipase (MGLL), alpha/beta-hydrolase domain containing protein 6 (ABHD6) and carboxylesterase 1 (CES1). These enzymes all show functional changes upon miR-185 overexpression.

As previously described, activity and abundance changes were validated using western-blotting pull-downs and RT-qPCR respectively. Although not a predicted target of miR-185, MGLL demonstrated one of the most significant activity decreases out of all the proteins not predicted to be direct targets, and showed the strongest mRNA downregulation (Fig. 1E, 2B, C). However, a 3’UTR luciferase reporter assay demonstrated the same decrease in luciferase signal with miR-185 as with our positive control, miR-182, a miRNA strongly predicted to bind the MGLL 3’UTR (Fig. S1C). It is therefore possible that the interaction between miR-185 and the MGLL 3’UTR is non-canonical in nature. ABHD6 was the only enzyme that demonstrated an increase in activity in our profiling with no accompanying change in its mRNA levels (Fig. 1E, 2C). Finally, our results show that miR-185 decreases CES1 activity by approximately a third (Fig. 1E). A similar miR-185 mediated decrease in CES1 mRNA levels was observed (Fig. 2C), indicating that CES1 activity is altered by transcriptional regulation. We have previously shown that CES1 siRNA knockdown leads to reduced HCV levels (Blais et al., 2010). In accordance with this, pharmacological inhibition with the WWL113 inhibitor showed up to a 50% decrease in HCV infection in a luciferase subgenomic replicon model (Fig. S1D). Therefore, the action of miR-185 on the CES1 enzyme is likely part of its anti-viral outcome. Together, the differential regulation of three different enzymes known to degrade 2-AG by miR-185 demonstrates a connection between miR-185 and the endocannabinoid pathway.

miRNA-185 downregulated serine hydrolases are putative PPAR-α targets

Overall, the decrease in the activities of these enzymes further confirms the role of miR-185 in modulating intracellular lipid metabolism while suggesting a common regulation mechanism. The peroxisome proliferator-activated receptors (PPARs) are nuclear transcription factors that act as central regulators of lipid metabolism in the cell and are well known to be differentially regulated during HCV infection (Dharancy et al., 2009). Interestingly, five of the aforementioned genes involved in lipid and 2-AG metabolism, AADAC, ACOT1, CES1, MGLL and LIPC, are putative PPAR-α targets and have been either observed or proposed to be regulated by the transcription factor (Dongol et al., 2007; Jones et al., 2013; Kersten and Stienstra, 2017). Since all of the putative PPAR-α target genes were seen to be significantly downregulated, we hypothesized that miR-185 acts as a co-regulator of this signal transduction pathway and may affect the levels of PPAR-α itself. RT-qPCR analysis showed that indeed, following miR-185 transfection, there was a significant decrease in the mRNA levels of the receptor (Fig. 2D). Interestingly, PPAR-α was not reliably predicted to be a target of miR-185 with bioinformatics. Thus, its perturbed function would not be predicted, highlighting the importance of using ABPP to determine functional effects of the miRNA.

Lipidomic analysis of microRNA function

To further cement the role of miR-185 in lipid metabolism, we set out to understand the changes to lipid species brought about by the microRNA. We expected variations in the major lipid classes known to affect the hepatitis C virus, as well as links to previously identified direct and indirect targets of the microRNA. Shotgun lipidomic analysis was performed on Huh7.5 cells overexpressing miR-185. As shown in figure 3, the miRNA causes a significant decrease in cholesteryl esters (CEs) and triacylglycerides (TAGs). This is consistent with the previously observed antiviral role of miR-185 against HCV via modulation of lipid membranes and droplets (Singaravelu et al., 2015). Interestingly, cholesteryl esters and triacylglycerides are the main components at the core of VLDL lipoprotein particles whose assembly pathway is intricately connected to the formation and release of HCV viral particles (Huang et al., 2007; Jones and McLauchlan, 2010).

Figure 3.

miR-185 changes the lipidomic profile of Huh7.5 cells. Cells were transfected with 100 nM of miR-185 for 72 hours, harvested and sent for shotgun lipidomics at Lipotype GmbH. A) Ratio of lipid mol % in treated vs control cells shows a significant decrease in cholesteryl esters and triglycerides, as well as a modest but significant increase in phosphatidylinositols. Error bars represent standard deviation, n=3. Significance assessed with two-tailed, unpaired student’s t-test, * = p<0.05. B) Heat map showing the mol % ratio of lipids in each category based on their total number of double bonds in the acyl chains. Grey squares denote where no data was obtained. Fold changes for each lipid species by category can be found in Fig. S2.

Unsaturation of lipids also plays an important role in both cellular membrane fluidity and maturation. We observed changes in lipid species based on the saturation of their acyl chains (Fig. S3A). Within triglycerides and phosphatidylglycerols we detected a decrease in saturated acyl chains, while species with more double bonds were seen to increase. In contrast, phosphatidates, phosphatidylethanolamines and phosphatidylserines showed a shift towards more saturated chains. This data suggests that microRNA-185’s antiviral effect is partly elicited via changes in lipid chain saturation.

To better understand how these changes to the lipid profile occur, we overlaid previous knowledge of miR-185 function with new information generated through our microarray and proteomic analysis. Figure 4 summarizes the enzymes involved in the synthesis of the major cellular lipid classes to visualize the changes that this small RNA induces in metabolic pathways. Fatty acid synthesis is at the basis of triacylglycerol and phospholipid synthesis and therefore, downregulation of FASN by miR-185 has a large impact on downstream products. In addition, numerous enzymes involved in lipid biosynthesis such as fatty acid elongases (ELOVL2–6) and acyl transferases (GPAT, AGPAT) were seen to decrease in transcript abundance via microarray (Table S3). These changes likely account for the significant reduction in TAGs, while downregulation of lipases such as MGLL and CES1 may explain the small increase in diacylglycerides (DAGs). Overall, the observed lipid phenotypes result from a complex combination of factors and changes in signaling cascades brought about by the miR-185’s effect on multiple transcripts.

Figure 4.

miR-185 regulates lipid metabolic pathways. miR-185 upregulated enzymes are in green while downregulated enzymes are in red. Enzymes in italics were present in the ABPP screen. Fold changes can be found in Tables S1, S2 and S3. CE: cholesteryl ester, TAG: triacylglycerol, PC: phosphatidylcholine, PI: phosphatidylinositol, DAG: diacylglycerol, PA: phosphatidate, PE: phosphatidylethanolamine, PG: phosphatidylglycerol, PS: phosphatidylserine.

Inhibition of MGLL reduces hepatitis C virus levels

Serine hydrolases are known to be good pharmacological targets, with drugs indicated for diseases such as obesity, diabetes and Alzheimer’s being prescribed everyday (Bachovchin and Cravatt, 2012). Therefore, we sought to determine if the antiviral effect of miR-185 could be recreated by targeting one of its principal serine hydrolase targets. Using a pharmacological inhibitor of MGLL, MJN110 (Niphakis et al., 2013), on Huh7.5 cells infected with the JFH1 2a strain of HCV resulted in a potent antiviral effect (Fig. 5A). However, this effect was not seen when the compound was used in cells harbouring the HCV luciferase subgenomic replicon model (data not shown). MGLL hydrolyses 2-AG into glycerol and arachidonic acid. We hypothesised that the antiviral effect may be due to an increase in 2-AG levels acting upon the CB1 pathway. Activation of the CB1 receptor by 2-AG is known to decrease cAMP production and therefore lower downstream targets such as phosphorylated AMPK (pAMPK), an inhibitor of SREBP1 (Ibsen et al., 2017; Mihaylova and Shaw, 2011). Sustained activation of CB1 has also been shown to result in receptor desensitization and functional antagonism of the pathway (Gainetdinov et al., 2004; Martin et al., 2004; Schlosburg et al., 2010), ultimately resulting in a downregulation of SREBP1. However, treatment with 1 μM MJN110 did not significantly affect pAMPK nor SREBP1 levels (Fig. 5B), suggesting a minimal effect on CB1. MGLL inhibition by MJN110 in the liver may exert its antiviral effects through reduced hydrolysis of 2-AG and other monoacylglycerols leading to broader effects on lipid homeostasis. Consistent with this hypothesis, an analysis of monoacylglycerol (MAG) and free fatty acid (FFA) species after a 72 hours treatment with the drug revealed a significant elevation in 18:1 MAG in both MJN110-treated and miR-185-treated cells, as well as a trend toward similar increases in 2-AG (C20:4 MAG) that did not achieve statistical significance (Fig. S3B).

Figure 5.

The miR-185 target MGLL is part of miR-185’s antiviral effect A) The MGLL inhibitor MJN110 inhibits HCV JFH1 infection in a dose-dependent manner with an EC₅₀ of 34nM, as measured by RT-qPCR of the HCV internal ribosome entry site (IRES). Cells were infected with HCV JFH1 for 5 hours after which the media was changed for fresh media containing MJN110, followed by a 72 hours incubation. B) Treatment with 1μM MJN110 does not result in a change in phospho-AMPK or SREBP1 levels. miR-185 transfection does not change phospho-AMPK levels, but does affect SREBP1 levels, as expected. AICAR is used as a positive control for AMPK phosphorylation. Both MJN110 and miR-185 lead to a decrease in MGLL levels. C) siRNA-mediated knock-down of MGLL, but not ABHD6 results in a significant downregulation of HCV JFH1 mRNA, comparable to the effects of MJN110. Cells were transfected with siRNA and incubated for 48 hours before being infected with HCV and incubated for another 72 hours. HCV JFH1 mRNA was measured by RT-qPCR, n=3. Significance assessed with two-tailed, unpaired student’s t-test, * = p<0.05, ** = p<0.01.

We tested the selectivity of MJN110 using competitive ABPP with the FP probe and confirmed that, as previously described (Niphakis et al., 2013), MJN110 is highly selective for MGLL with the exception of a partial inhibition of ABHD6 (Fig. S4). Interestingly, we observed a substantial decrease in MGLL protein abundance with the drug, with no change to ABHD6 levels (Fig. S4B), suggesting potential degradation of the drug-enzyme complex. To our knowledge, degradation of MGLL by MJN110 has not been previously reported. We further confirmed that 2-AG levels were affected by treatment of Huh7.5 cells with MJN110. After 4 and 24 hours of treatment, levels of 2-AG increased approximately 7-fold and increased by 3-fold after 72 hours (Fig. S3C), supporting the hypothesis that 2-AG levels are regulated significantly by MGLL activity in human hepatoma cells. Unexpectedly, arachidonic acid levels were also seen to increase, potentially as a feedback response to treatment. In order to determine if ABHD6 was contributing to MJN110’s effect on HCV, we used small interfering RNAs to specifically knock-down MGLL, ABHD6 or both enzymes. Infection of MGLL siRNA-transfected cells with HCV JFH1 lead to a significant reduction in viral RNA after a 72 hours incubation, similar to that seen with MJN110 (Fig. 5C). Knock-down of ABHD6 alone as well as simultaneous knock-down of both the enzymes did not show reduction in viral load. Interestingly, these effects were not as pronounced when the infected cells were only incubated for 48 hours (Fig. S5A). Successful reduction in MGLL and ABHD6 protein levels were confirmed via western blotting at 48, 96 and 120 hours post-transfection (Fig. S5B) and a time-dependent reduction in MGLL levels with MJN110 was once again observed up to 96 hours post-treatment (Fig. S5C). Overall, downregulation of MGLL by miR-185 in the liver may be an important part of its antiviral effect.

Discussion

The interest in miR-185 stems from its well-known function in regulating metabolism (Wang et al., 2013, 2014; Yang et al., 2014), and its recently identified antiviral activity against hepatitis C and other viruses (Singaravelu et al., 2015). However, the mechanisms through which the microRNA exerts its effects are complex and not fully understood. We sought to investigate the implication of metabolic serine hydrolases in miR-185’s reshaping of the cellular lipid profile through activity-based protein profiling, supplemented with transcriptomic and lipidomic analysis. Our findings not only identified new targets of the microRNA, but also demonstrate that ABPP can report on enzyme behavior which would be otherwise missed with traditional transcriptomics analysis.

Firstly, we identified FASN, AADAC, ACOT1 and LIPC as lipid metabolic enzymes regulated by miR-185. FASN catalyzes the de novo synthesis of fatty acids (Maier et al., 2008) and has been shown to increase in activity during HCV infection (Nasheri et al., 2013). It is clear that interference in de novo fatty acid synthesis via down-regulation of FASN expression represents one of the mechanisms by which miR-185 decreases lipid abundance. Overexpression of AADAC, an enzyme responsible for triglyceride hydrolysis, in hepatoma cells has been shown to reduce triacylglycerol levels and increase fatty acid oxidation (Lo et al., 2010). The enzyme has also been shown to be involved the mobilization of triacylglycerols for the production of VLDL precursors and its knock-down impaired HCV infectivity (Nourbakhsh et al., 2013). Acyl-CoA thioesterase 1 (ACOT1) hydrolyses acyl-CoAs to free fatty acids and CoA, and is considered an auxiliary enzyme in β-oxidation (Hunt et al., 2014). Modulation of ACOT1 by miR-185 could potentially decrease lipid levels by affecting β-oxidation. LIPC, also known as hepatic lipase, is able to hydrolyze triglycerides from chylomicrons, catalyze the conversion between lipoproteins of different densities and facilitate uptake of lipoprotein remnants (Chang and Borensztajn, 1993). Hepatotropic viruses such as HCV interact significantly with lipoprotein pathways during their life cycle and some studies have shown that the activity of LIPC may have an anti-viral role by disrupting those interactions (Shimizu et al., 2010; Shinohara et al., 2013). Interestingly, we observed a greater reduction in LIPC activity than abundance. It is known that this enzyme resides in the ER in association with calnexin and requires post-translational modifications to be secreted in its active form (Ben-Zeev and Doolittle, 2004). Our data suggests that the activity of this enzyme is decreased by miR-185, demonstrating a functional connection between enzyme activity and metabolic and immunometabolic function.

In assessing the changes in enzyme activity induced by miR-185, we discovered a possible link between miRNA signaling and the endocannabinoid pathway. Specifically, three enzymes (MGLL, ABHD6 and CES1) identified in our screen are known to hydrolyze 2-AG, one of the two main endocannabinoid lipid species. Of these enzymes, MGLL and CES1 were downregulated by miR-185 while ABHD6 showed a modest increase in activity. MGLL catalyzes the conversion of monoacylglycerides to free fatty acids and glycerol and is responsible for the majority of 2-AG degradation in the brain (Savinainen et al., 2012). Interestingly, mice deficient in the enzyme have been shown to exhibit high liver triacylglycerol levels and reduced VLDL secretion (Taschler et al., 2011) and expression of MGLL mRNA was shown to be decreased in liver biopsies of chronic HCV patients (Patsenker et al., 2015). ABHD6 has been shown to degrade 2-AG in the brain (Marrs et al., 2010); however, the function of the enzyme in other tissues and systems is not well understood. Interestingly, Thomas, G. et al. have shown that ABHD6 is an important modulator of the metabolic syndrome and that knockdown of the enzyme protected mice from high-fat-diet-induced obesity and hepatic steatosis. Furthermore, ABHD6 knockdown increased expression of lipolytic genes such as hormone-sensitive lipase and MGLL (Thomas et al., 2013). It is therefore plausible that a feedback mechanism exists between ABHD6 and MGLL to maintain 2-AG levels. CES1, also known as triacylglycerol hydrolase (TGH), plays an important role in lipid metabolism (Ross et al., 2010) and a 2010 study has shown that CES1 is also capable of hydrolyzing 2-AG (Xie et al., 2010). It is thought that although most 2-AG degradation in the brain occurs through MGLL and ABHD6, CES1 may play a role in the endocannabinoid pathway in peripheral tissues (Xie et al., 2010). Inhibition of the enzyme decreased apoB-100 secretion in primary rat hepatocytes and altered VLDL production (Gilham et al., 2003). It is therefore not surprising that the enzyme was shown to increase in activity with HCV infection (Blais et al., 2010) and that our results show an antiviral effects with inhibition of the enzyme. Overall, since the endocannabinoid pathway has been shown to be intricately linked to lipid metabolism and signaling, miR-185 may play a more complex cellular role than previously thought.

When taken together, it is the simultaneous changes in the activities of all these serine hydrolases, as well as other non-enzymatic factors previously described (Singaravelu et al., 2015) which most likely lead to an antiviral lipid-poor cellular environment. We further show that microRNA-185 decreases levels of the nuclear receptor PPAR-α. This is the most highly expressed PPAR in the liver (Auboeuf et al., 1997), and it responds to intracellular levels of free fatty acids to regulate β-oxidation, triglyceride hydrolysis, lipogenesis and ketone body synthesis. We previously showed that PPAR-α antagonism has antiviral effects on the hepatitis C virus by altering lipid metabolism in the cell (Rakic et al., 2006). It is also noteworthy that ACOT1 and CES1 were shown to regulate PPAR-α activity by producing fatty acid ligands for the receptor (Franklin et al., 2017; Xu et al., 2014). Thus miR-185’s immunometabolic function and ultimate antiviral effects are connected with co-regulation of PPAR signaling through functional targeting of lipid metabolism.

To gain greater insight into miR-185’s effects on lipid pools, we performed shotgun lipidomic analysis and observed a decreased in cholesteryl esters and triacylglycerides. We linked these changes to the enzymes identified via ABPP as well as other miR-185 targets found in our microarray analysis and summarized them in figure 4. Part of the miR-185-altered phenotype is thought to be achieved through a decrease in the transcripts of regulatory genes such as SREBPs and PPARs. SREBP2, previously shown to be downregulated by miR-185 (Singaravelu et al., 2015), is the main transcription factor responsible for the regulation of cholesterol biosynthesis. Cholesterol is acylated into cholesteryl esters via two ways, the ACAT and LCAT pathways. Although microarray analysis (Table S2) showed an upregulation of ACAT and a downregulation of LCAT, our ABPP profiling results (Fig. 1E) indicated a decrease in the activity of CES1, which is able to hydrolyze cholesteryl esters. Though this might be expected to increase total CE levels, the overall decrease in CE indicates that these changes may be eclipsed by the reduction in cholesterol substrate brought about by the downregulation of SREBP2.

TAGs and CEs are incorporated into the core of VLDL particles for release from hepatocytes into the circulation. HCV is well known for using the VLDL pathway for its own secretion (Huang et al., 2007). The virus has also been shown to increase CE synthesis through modulation of ACAT genes, and inhibition of cholesterol esterification is known to be antiviral (Loizides-Mangold et al., 2014; Read et al., 2014). Therefore, by decreasing the levels of TAGs and CEs, miR-185 could be reducing the release of viral particles. VLDL particles are remodeled into intermediate density IDL particles by lipoprotein lipase and into LDL particles by LIPC. We observed a small increase in LPL abundance but a significant decrease in the activity, but not the abundance, of LIPC. This suggests that miR-185 is not only affecting the internal lipid environment but also remodeling the profile of lipoproteins secreted externally.

Interestingly, we also observed a significant increase in phosphatidylinositols, a signaling lipid species known to be enriched in arachidonic acid (Barneda et al., 2019). This may be brought about by miR-185’s modulation of the enzymes MGLL, ABHD6 and CES1 involved in endocannabinoid lipid metabolism as previously described. However, miR-185’s effect on this pathway remain to be investigated.

To better understand how the microRNA affects lipid-rich structures, we focused our attention on the saturation levels of each lipid species. We observed a decrease in saturated triglycerides and phosphatidylglycerols and an increase in saturated phosphatidates, phosphatidylethanolamines and phosphatidylserines. These changes may be part of the miR-185-induced remodeling of lipid membranes compositions which occurs during the immune response to HCV infection. Indeed, many RNA viruses are known to prefer unsaturated fatty acids and to form modified ER-membrane compartments where viral replication occurs. miR-185 was previously shown to downregulate stearoyl-CoA desaturase 1 (SCD-1), one of the principal enzymes in fatty acid desaturation (Lyn et al., 2014). We also observed a decrease in the fatty acid desaturases 1 and 2. Overall, these changes in the phospholipids may affect chain flexibility and membrane fluidity as well as curvature, contributing to miR-185’s antiviral effect. Taken together, there is a strong correlation between changes in enzyme activity, transcript abundance and downstream changes in lipid content mediated by miR-185.

Finally, we show that pharmacological inhibition of MGLL by the inhibitor MJN110, as well as siRNA-mediated knock-down of the enzyme, results in a potent downregulation of HCV JFH-1 infection. We confirmed that knock-down of ABHD6, a potential off target of MJN110, does not lead to a reduction in infection, validating that the drug is acting through its primary target. Although we observed an increase in MGLL’s substrate, 2-AG, there was no effect on genes downstream of the CB1 receptor, suggesting that the antiviral effect may not be mediated through this pathway. Therefore, other possibilities likely contribute to the observed antiviral effect. MGLL is responsible for the hydrolysis of a large variety of monoacylglycerols which affects free fatty acid (FFA) pools. Nomura et al. have shown that increased MGLL activity significantly affects cellular FFA levels and promote carcinogenicity (Nomura et al., 2010), and the link between endogenous fatty acids and HCV pathogenesis is well established (Bassendine et al., 2013; Kapadia and Chisari, 2005). Although we did not observe significant changes in palmitic, stearic and oleic acid levels, MJN110 did significantly increase the levels of the 18:1 oleic acid monoacylglycerol, a lipid previously shown to prevent LDL oxidation (Cho et al., 2010). Interestingly, transfection with miR-185 also led to an increase in this species, as well as in 2-AG and arachidonic acid. It is therefore possible that MJN110 acts by creating an unfavorable lipid turnover environment for the virus. Additionally, arachidonic acid is at the base of prostaglandin synthesis through the action of cyclooxygenase enzymes, and prostaglandin A₁ (PGA₁) has been show to have antiviral potential against HCV (Tsukimoto et al., 2015). Finally, the antiviral effect of MJN110 was not observed in a subgenomic replicon model of HCV, suggesting the metabolic changes may affect cell entry or release, as opposed to replication. This is corroborated by the observation that MGLL deficient mice have reduced VLDL secretion (Taschler et al., 2011). Future work should be directed at better understanding the link between this enzyme and HCV.

In this work, using a combination of ABPP, transcriptomics and lipidomics, we determined new functional roles for miR-185 by targeting a sub-population of the proteome. We confirmed miR-185’s role in lipid metabolism via direct and indirect enzymatic modulation, as well as a more intricate involvement of this microRNA in lipid signaling pathways. Our study has shown that miR-185 affects many more targets functionally than previously thought, both directly and indirectly. Its role in lipid turnover and shuttling is furthered by its regulation of PPAR-α, thereby influencing the expression of downstream lipolytic genes. Generally, we have demonstrated that miR-185 regulates the levels of total lipid pools. All these changes taken together improve our understanding of the miR-185’s metabolic and anti-viral effects. Additionally, our results clearly demonstrate the utility of ABPP in uncovering new direct and indirect targets of non-coding RNAs. Here we show that a systematic analysis of the functional proteome can provide a global view of the influence of miR-185 on lipid metabolism. Furthermore, we demonstrate that it is possible to achieve pharmacological recapitulation of miRNA phenotype using inhibitors of the protein targets regulated by the microRNA. Since microRNA therapeutics are challenging to produce and deliver, our study points to a path for replicating microRNA phenotypes with small molecules.

STAR METHODS

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, John Paul Pezacki (john.pezacki@uottawa.ca).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

Microarray data was obtained from Singaravelu et al.(Singaravelu et al., 2015). Data files are available on NCBI GEO depository (GSE73165). Mass spectrometry raw data for the activity-based protein profiling experiment is available in Mendeley data (DOI:10.17632/2h8mf2k3kt.1).

Experimental Model and Subject Details

The human hepatoma Huh7.5 cell line (sex: male) was derived from the Huh7 line for the purpose of permissive HCV infection (Blight et al., 2002). The line was a kind gift from Dr. C.M. Rice (Rockefeller University). Huh7 cells harboring the luciferase sub genomic replicon (pFK-I389neo/luc/NS3–3’/5.1) were previously generated in our group (Sagan et al., 2006). All cells were grown at 37°C, 5% CO₂ in high glucose DMEM (Gibco, Life Technologies) with 10% FBS (Wisent) and 10 mM non-essential amino acids (ThermoFisher Scientific).

Method details

miRNA transfection and lysis for active proteome labeling

hsa-miRNA-185-5p miRNA mimic or control miRNA (miRVana, ThermoFisher Scientific) were combined with Lipofectamine RNAiMAX transfection reagent (ThermoFisher Scientific) in Optimem (Gibco, Life Technologies) medium as previously described (Singaravelu et al., 2015). The mixtures were added to the growth media of cells for a final concentration of 100 nM of miRNA. Cells were lysed after 72 hours. Confluent cells were washed with sodium phosphate buffer saline (PBS) and harvested with 1% Triton-X 100 in 10 mM PBS. Cells were lysed by sonication (15 one second pulses), centrifuged at 20000g, 4°C for 5 min, and the supernatant was stored at −80°C until labelling. Protein content was quantified using a DC protein assay (Bio-Rad).

Active proteome labeling with fluorophosphonate-TAMRA

Huh7.5 cells were cultured, transfected with miR-185 and lysed as described above. The lysates were diluted at 1mg/ml in 50ul of lysis buffer and labelled with FP-TAMRA (Thermo Fisher) at 1uM for 1hour at 37°C. “No probe” samples were incubated with the equivalent amount of DMSO only. Heat denatured samples were heated for 15 minutes at 95°C before labelling. The reaction was quenched by the addition of SDS gel loading buffer and the samples were run on a large 10% SDS-PAGE gel. The gel was imaged for rhodamine fluorescence using the ChemiDoc MP system (Bio-Rad) after which it was stained for whole protein with Coomasie stain, destained and re-imaged.

Active proteome labeling with fluorophosphonate-biotin

ABPP labelling of serine hydrolases with FP-biotin was performed as described by Barglow and Cravatt (Barglow and Cravatt, 2007). Protein lysates were diluted in lysis buffer to 1 ml of 1 mg/ml for western-blots pull downs and 2 mg/ml for mass-spectrometry. Fluorophosphonate-biotin probe (Santa Cruz Biotechnologies) or DMSO negative control was added to active proteome for a final concentration of 5 μM and incubated with rotation for 1 hour at 37°C. Proteins were precipitated in 5X volume of acetone for at least 15 minutes at −80°C and centrifuged 5 minutes at 6500 g and 4°C. The acetone was removed and the protein pellets were washed with 750 μl cold methanol in 3 alternating cycles of sonication (5 one second pulses) and centrifugation at 6500 g for 5 minutes. The proteins pellets were then dissolved in 650 μl 2.5% SDS in PBS and sonicated for 15 pulses. The samples were then heated for 5 minutes at 60°C followed by centrifugation at 6500 g for 4 minutes to pellet any remaining contaminants. Aliquots of the supernatants were taken for input control after which the samples were mixed with PBS for a final volume of 8 ml.

Streptavidin enrichment

100 μl of streptavidin-agarose beads (ThermoFisher Scientific) were washed with PBS in biospin columns and added to the protein samples, followed by incubation with rotation for 1.5 h at room temperature. The beads were pelleted by centrifugation at 1400 g for 2 minutes, transferred into biospin columns (Bio-Rad) and washed 3 times with 1% SDS followed by three washes with 6 M urea.

Immunoblotting

Streptavidin beads were washed 3 times with PBS and incubated with 2X SDS-PAGE loading buffer at 95°C for 15 minutes. The resulting eluates were run on a 10% TGX Stain-Free^™ FastCast^™ Acrylamide gel (Bio-Rad). Proteins were transferred from the gel to a PVDF membrane using the Trans-Blot® Turbo^™ Transfer System (Bio-Rad), blocked with 5% milk in TBS-Tween for 1 hour at room temperature and probed with the appropriate primary and secondary antibody dilutions. Imaging for chemiluminescence was performed using the ChemiDoc MP system (Bio-Rad).

Proteomic mass-spectrometry sample preparation

Streptavidin beads were washed once with PBS and 5 times with 50 mM ammonium bicarbonate (ABC), pelleted by centrifugation at 1400 g for 2 minutes, then resuspended in 500 μl of 10 mM DTT in ABC. Samples were heated at 65°C for 15 min and iodoacetamide was added for a final concentration of 25 mM. Samples were rotated at room temperature for 30 min in the dark. Beads were centrifuged and washed with 50 mM triethylammonium bicarbonate (TEAB), pH 8.5. Proteins were digested off the beads overnight with 10 μg/ml trypsin in TEAB at 37°C. Beads were pelleted, the supernatant collected and the protein vacuum dried before dimethyl labelling.

Dimethyl labelling and desalting

Dimethyl labelling was performed based on the method of Boersema et al (Boersema et al., 2009). Digested samples were reconstituted in 100 μL of 100 mM TEAB to which 4 μL of 4% (vol/vol) CH₂O (light), CD₂O (medium) or ¹³CD₂O (heavy) were added. Untransfected lysates treated with DMSO were subjected to the “light” condition, control miRNA transfected and FP-biotin-labelled lysates were subjected to the “medium” condition, while miR-185 transfected and FP-biotin-labelled lysates were subjected to the “heavy” condition. Afterwards, 4 μL of 0.6 M NaBH₃CN were added to the light samples, and 4 μL of 0.6 M NaBD₃CN were added to the heavy sample. The samples were incubated for 1 h at room temperature while rotating after which, 16 μL of 1% (vol/vol) ammonia solution and 30 μL of 5% formic acid were mixed in to quench the reaction and acidify the samples. The differentially labelled samples were mixed 1:1, desalted using Pierce C18 spin columns (Thermo Fisher) and vacuum dried.

Proteomic mass-spectrometry and data analysis

MS analysis was performed by Dr. Gleb G. Mirinov, John L. Holmes Mass Spectrometry Facility, Department of Chemistry and Biomolecular Sciences, University of Ottawa. Digested peptides were analyzed by HPLC-MS/MS using a Dionex Ultimate 3000 nano-HPLC system with an Acclaim PepMap RSLC 75 μm ID × 150 mm length separation column (Thermo Scientific, San Jose, CA), coupled with Orbitrap Fusion mass spectrometer (Thermo Scientific, San Jose, CA). 5 μl of sample were injected and separated by the following gradient (solution A: 0.1% formic acid in H₂O; solution B: 80% acetonitrile, 0.1% formic acid in H₂0) with the flow of 200 nl/min, such that 0.0–80.0 min increased between 0–40% B, 80.0–80.1 min rose to 40–80% B, 80.1–90.0 min rose to 80% B, 90.0–90.1 min decreased between 80–2% B, and 90.1–115.0 min maintained 2% B. Peptides were ionized using nano-ESI with spray voltage in positive mode at 2000 V. Ion transfer tube temperature was 275°C and the S-lens RF level was 60. Survey scans were performed on peptide precursors between 300 and 1500 m/z at 60K resolution (at 200 m/z). The ion count target was set to 2 × 10⁵ and the maximum injection time 50 ms. Precursor peptides for tandem MS analyses were isolated by quadrupole at 0.7 Th. CID fragmentation was performed with collision energy of 35% and 5% step, and Normal scan MS analysis in the ion trap. The MS² ion count target was set to 10⁴ and the max injection time was 35 ms. Precursors with charge state 2–6 were sampled for MS². The dynamic exclusion duration was set to 60 s with 10 ppm tolerance around a precursor ion and its isotopes. The instrument was run in 4 s cycles in top speed mode. Proteome Discoverer (version 1.4.1.14, Thermo Scientific, San Jose, CA) was used to process the raw data, and MS² spectra were searched against a UniProt database for Homo sapiens (Human) (http://www.uniprot.org) with SEQUEST HT engine. Peptides were generated from a tryptic digestion containing up to two missed cleavages. Fixed modifications consisted of carbamidomethylation of cysteines, and variable modifications consisted of oxidation of methionines and protein N-terminal acetylation. Precursor mass tolerance was 10 ppm and product ions were searched at 0.6 Da tolerances. A target decoy validation with FDR of 1% was used to validate peptide spectral matches (PSM). Protein intensities from treated samples were normalized over control transfected samples.

Quantitative PCR

RNA isolation was performed using the RNeasy (Qiagen) isolation kit as per manufacturer’s protocol. RNA integrity was confirmed by electrophoresis on 0.8% agarose gel in 1× TBE (Ambion). Reverse transcription was performed using the iScript cDNA Synthesis Kit (Bio-Rad) using 500 ng of RNA as per manufacturer’s protocol. Quantitative PCR (qPCR) was performed using SYBR Green Supermix (Bio-Rad) as per manufacturer’s protocol on the CFX Real-Time PCR Detection System (Bio-Rad) with the primer sequences are listed in Supplemental Table 4. 18S rRNA was used for normalisation and expression fold changes relative to mock treatments were calculated using the 2^−ΔΔCt method.

HCV subgenomic replicon luciferase assays

Huh7 cells constitutively expressing the HCV sub-genomic replicon linked to a luciferase reporter gene (pFK-I389neo/luc/NS3–3’/5.1) were seeded in 24 well plates at 20000 cells/well. Cells were treated with a serial dilution of WWL113 (Sigma-Aldrich) for final concentrations of 0.024 μM to 25 μM and 1% DMSO for 24 or 72 hours, washed with PBS and lysed with passive lysis buffer (Promega). Luciferase assay was performed in technical triplicates in a 96 well-plate as previously described (Dyer et al., 2000). Luciferase signal was normalized over protein concentration and mock DMSO treatment.

MJN110 and Fluorophosphonate competitive ABPP

Huh7.5 cells were cultured as described above and one day before treatment, were seeded in 6 well plates at 60000 cells/well. The next day, cells were treated with 1 μM MJN110 or DMSO control. After 72 hours, the cells were harvested with 1% Triton-X 100 in 10 mM PBS and lysed as described above. The samples were diluted at 0.75 ug/μl in 70 μl for FP-TAMRA labelling and 1.5 mg/ml in 1 ml for FP-Biotin. FP-TAMRA (Thermo Fisher) was added for a final concentration of 0.7 μM and incubated at 37°C for one hour, after which the proteins were precipitated with 1 ml acetone for at least 15 minutes at −80°C and centrifuged 5 minutes at 14000 g and 4°C. The acetone was removed, and the protein pellets were dried and dissolved in 2X SDS-PAGE loading buffer before being loaded on a 10% TGX Stain-Free^™ FastCast^™ Acrylamide gel (Bio-Rad). The gel was imaged for rhodamine fluorescence using the ChemiDoc MP system (Bio-Rad) after which proteins were transferred to a PVDF membrane and immunoblotting for MGLL and ABHD6 was performed as described above. Labelling with FP-Biotin, streptavidin enrichment, pull-down and immunoblotting were performed as previously described. For the higher resolution gel, labelling of MJN110-treated lysates and gel electrophoresis were performed as previously described for the miR-185-transfected cell lysates.

Samples preparation for analysis of 2-AG and AA levels

Huh7.5 cells were seeded one day prior to treatment in full media (DMEM, 10% FBS, NEAAs) at 60000 cells/well in 6 well plates. Cells were treated with 1 μM MJN110 or DMSO control for 4 hrs, 24 hrs or 72 hrs. Lipid extraction was performed following an acidified Bligh and Dyer method as follows. Cells were washed, trypsinized and lysed in 500 μl cold methanol containing 100nM arachidonic acid-d8 (d8-AA) spike-in (Cayman) to which 1.5 ml of MeOH, 1 ml chloroform and 0.8 ml water were added. The samples were shaken at room temperature at 700 rpm for 1 hour after which 1 ml of water, 50 μl formic acid and 1 ml chloroform were added. The samples were then vortexed and spun at 2000 g and 4 °C for 5 min. The organic layer was separated and vacuum dried for mass-spectrometry. The remainder fractions were combined with 400 μl MeOH and 600 μl chloroform, vortexed and centrifuged at 2000g for 5min. The top aqueous layer was removed and 2400 μl of MeOH were added. The samples were vortexed and centrifuged at 4500g for 5 min at 4°C after which the supernatants were removed, and the pellets air-dried. The protein pellets were then resuspended in 2X SD-SPAGE loading buffer before being loaded on a 10% TGX Stain-Free^™ FastCast^™ Acrylamide gel (Bio-Rad) for total protein quantification and immunoblotting.

Lipidomic analysis

Huh7.5 cells were cultured and transfected as described above. Confluent cells were trypsinized, spun down and resuspended in Ca²⁺, Mg²⁺-free PBS. Cells were counted and resuspended in PBS before being sent for analysis at Lipotype GmbH. The amounts of the lipid classes were normalized to the total lipid amount to obtain mol % per total lipids. Mol % of the treated samples were divided by their respective control to obtain the mol % ratio.

Microarray analysis

Microarray data was obtained from Singaravelu et al.(Singaravelu et al., 2015). Data files are available on NCBI GEO depository (GSE73165). Briefly, Huh7.5 cells were cultured and transfected as described above. RNA isolation was performed using the RNeasy (Qiagen) isolation kit as per manufacturer’s protocol. Expression profiling was performed in duplicate using Affymetrix Human Gene ST.2.0 arrays. Normalized and analysis was performed using the Affymetrix Expression Console and Transcriptome Analysis Console.

3’UTR luciferase reporter analysis

Huh7.5 cells were seeded in 6 replicates at 40000 cells/well in a 24 well plate. The following day, cells were transfected with 200 ng/well MGLL 3’UTR pEZX-MT06 plasmid (GeneCopoeia) using Lipofectamine 2000 transfection reagent as per manufacturer’s protocol (Thermofisher Scientific). After 24 hours, cells were transfected with 100 μM control miRNA, miR-185 or miR-182 (miRVana, ThermoFisher Scientific) using Lipofectamine RNAiMAX transfection reagent (ThermoFisher Scientific) as per manufacturer’s protocol. Cells were incubated for 48 hours before being lysed with passive lysis buffer (Promega). Luciferase assay was performed in technical triplicates in a 96 well-plate as previously described (Dyer et al., 2000).

HCV JFH1 infection and MJN110 treatment

Huh7.5 cells were infected with the HCV JFH1_T strain described by Russell et al. (Russell et al., 2008). Cells were seeded one day prior to infection in full media (DMEM, 10% FBS, NEAAs) at 20000 cells/well in 24 well plates. The next day, the cells were incubated with the virus at MOI 0.1 in serum-free DMEM for 5 hours, after which the infection media was replaced with full media containing either 1% DMSO or MJN110 at different dilutions in DMSO. Final MJN110 concentrations ranged from 6.0E-6 μM to 3.1 μM. The cells were incubated for 72 hours after which they were lysed in RL buffer and RNA was isolated using the Norgen Single Cell RNA purification kit (Norgen Biotek) as per manufacturer’s protocol. Reverse transcription and qPCR were performed as described above with primers targeting the HCV JFH1 IRES (sequences listed in Supplemental Table 4).

Relative analysis of MAG and FFA species

Huh7.5 cells were seeded at 1×10⁶ cells for miR-185 transfection and at 750000 cells for MJN100 treatment in 100mm dishes. Transfection and treatments were performed as described above (100nM miR-185 and 1μM MJN110). After 72 hours, the cells were washed with PBS, scraped, pelleted and flash frozen in liquid nitrogen. Cell pellets were thawed on ice and the total cell metabolome was extracted in 4 mL 2:1:1 CHCl₃/MeOH/DPBS (v/v/v) solution containing the internal standard mix (500 pmol 2-AG-d5 and 1 nmol AA-d8). The mixture was vortexed vigorously and centrifuged at 2000g for 5 min at 4 °C. The bottom organic phase was collected, and the remaining aqueous phase was acidified with 20 μL formic acid and re-extracted by the addition of 2 mL CHCl₃. Both of the organic phases were pooled, dried down under N2 gas, and reconstituted in 150 μL 2:1 CHCl₃/MeOH (v/v) for LC/MS analysis. Metabolites analyzed in this study were quantified using LC/MS–based multiple reaction monitoring (MRM) methods (Agilent Technologies 6470 Triple Quad). MS analysis was performed using ESI with the following parameters: drying gas temperature, 350 °C; drying gas flow, 9 l/min; nebulizer pressure, 45 Ψ; sheath gas temperature, 375 °C; sheath gas flow, 12 l/min; fragmentor voltage, 100 V; and capillary voltage, 3.5 kV. The MRM transitions for the targeted LC/MS analysis are presented in Supplemental Table 5. The separation of metabolites was achieved using a 50 mm × 4.6 mm 5 μm Gemini C18 column (Phenomenex) coupled to a guard column (Gemini: C18: 4 × 3 mm). For negative mode analysis, H₂O:MeOH (95:5, v/v) with 0.1 % NH₄OH (v/v) and iPrOH:MeOH:H₂O (60:35:5, v/v) with 0.1 % NH₄OH (v/v) were used as buffer A and B, respectively. For positive mode analysis, H₂O:MeOH (95:5, v/v) with 0.1 % formic acid (v/v) and iPrOH:MeOH:H₂O (60:35:5, v/v) with 0.1 % formic acid (v/v) were used as buffer A and B, respectively. The LC gradient for negative mode analysis was the following after injection: 20% B at 0.1 mL/min for 5 min; then increase to 85% B at 0.4 mL/min for 15 min; increase to 100% B at 0.5 mL/min for 5 min, run at 100% B at 0.5 mL/min for 2 min; then go back to 20% B and equilibrate at 0.5 mL/min for 5 min. The LC gradient for positive mode analysis was the following after injection: start from 100% A and increase to 60% B at 0.1 mL/min for 5 min; increase to 100% B at 0.4 mL/min for 15 minutes; maintain 100% B at 0.4 mL/min for 13 minutes and then go back to 100% A at 0.4 mL/min and equilibrate for 1 minute. Lipid species were quantified by measuring areas under the curve in comparison to the corresponding internal standards and then normalizing to the cellular protein concentration. Two batches of three biological replicates were performed. Lipid content for each sample was normalized over its total protein content and the results are reported as a percentage of the average control group of each sample respectively.

siRNA knock-down of MGLL and ABHD6 and HCV JFH1_T infection

Huh7.5 cells were seeded at 1.5×10⁵ cells per well in 6 well plates. The following day, siRNA mimic or control siRNA (ThermoFisher Scientific) were combined with 7 μl Lipofectamine RNAiMAX transfection reagent (ThermoFisher Scientific) in Optimem (Gibco, Life Technologies) medium for a final siRNA concentration of 50 nM in each well. The mixtures were added to the growth media of cells and incubated for 48 hours. For the MJN110 treatments, cells were seeded at 7.5×10⁴ at the same time. After 48 hours of incubation, the media was replaced with serum-free DMEM containing HCV JFH1_T at MOI of 0.1 and the cells were incubated for 5 hours. Following this, the infection media was replaced with full media. For the MJN110 treatments, 1μM MJN110 or DMSO was added at this time. The cells were incubated for either 48 or 72 hours after which they were lysed in RLT buffer and the RNA was isolated using the RNeasy (Qiagen) isolation kit as per manufacturer’s protocol. Reverse transcription and qPCR were performed as described above with primers targeting the HCV JFH1_T IRES.

Relative analysis of 2-AG and AA levels using mass-spectrometry

MS analysis was performed by Dr. Zoran Minic, John L. Holmes Mass Spectrometry Facility, Department of Chemistry and Biomolecular Sciences, University of Ottawa. Analysis of arachidonic acid and arachidonic acid-d8 were performed in negative mode. Samples were resuspended in 0.3 ml acetonitrile-water mixture (80:20, v/v) and vortexed for 3 min. The suspension was centrifuged for 30 s at 15000g and the supernatant was used for mass spectrometry analysis. Analysis of 2-arachidonyl glycerol was done in positive mode. Sample preparation was performed as for the negative-ion mode, with the exception that the solubilization solution was acidified with 0.1% formic acid. The obtained samples were analyzed by direct infusion electrospray ionisation (ESI), at a flow rate of 6 μl/min, in an Ion Max API source coupled to Q Exactive Plus MS (ThermoScientific, Bremen, Germany). The instrument was calibrated with Pierce ESI Positive and Negative Ion Calibration solution (Pierce #88323 and #88324). Measurements were carried out in the negative-ion mode with the experimental conditions applied as follows: the ion-spray potential of 3.4 kV and all other parameters (sheath gas flow rate 10, S-lens RF level 50, AGC target 3e6) optimized for maximum molecular ion transmission, and the transfer capillary temperature at 320 °C. For the positive-ion mode the experimental conditions were set as for the negative-ion mode with a difference that the ion-spray potential was 4.0 kV and sheath gas flow rate of 4. Full scan high resolution mass spectra (R=75000 at m/z 400) were collected at a selected m/z range with a maximum injection time of 100 ms. For data acquisition and data processing, Xcalibur software was used. To obtain stable signals, samples were first electrosprayed for 5 min and then processed at the acquisition time of 1 min. The selected m/z range for analysis of arachidonic acid, arachidonic acid d8 and 2-arachidonoylglycerol were 302–304, 311–313 and 378–381, respectively. The relative intensity of precursor ions at m/z values corresponding to the predicted ratio and previously validated with pure compound were used for relative quantification. Lipid intensities were normalized over the AA-d8 spike-in control and over the densitometric intensity of the protein fractions run on an SDS-PAGE gel.

Statistical analysis

Data analysis was performed in Microsoft Excel and GraphPad Prism software. Fold changes were calculated relative to mock transfected or treated samples. Significance was assessed with two-tailed, unpaired student’s t-test where p<0.05 was considered significant.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
PTP1D (SHP2)	BD transduction Labs	Cat# 610621; RRID: AB_397953
AADAC (G-12)	Santa-Cruz Biotechnologies	Cat# sc-390591
SCD1	Abcam	Cat# ab19862; RRID: AB_445179
FASN	Santa-Cruz Biotechnologies	Cat# sc-483570
MGLL	Santa-Cruz Biotechnologies	Cat# sc-398942
LIPC (hepatic lipase)	Santa-Cruz Biotechnologies	Cat# sc-21740; RRID: AB_627888
ACOT1/2	Santa-Cruz Biotechnologies	Cat# sc-373917; RRID: AB_10918465
CES1	Kind gift from Dr. Lance Pohl, NIH, Bethesda, MD, USA	N/A
ABHD6	Cell Signaling Technology	Cat# 97573S; RRID: AB_2800281
pAMPK (Thr172)	Cell Signaling Technology	Cat# 2531S; RRID: AB_330330
AMPK	Cell Signaling Technology	Cat# 2532S; RRID: AB_330331
SREBP1	Santa-Cruz Biotechnologies	Cat# Sc-366; RRID: AB_2194229
HRP goat α-mouse	Jackson Immuno-research	Cat# 115-035-062; RRID: AB_2338504
HRP donkey α-rabbit	Jackson Immuno-research	Cat# 711-035-152; RRID: AB_10015282
Bacterial and Virus Strains
HCV JFH1T	Kind gift from Dr. Rodney Russell, Memorial University, NFL, CA (Russell et al., 2008)	N/A
Chemicals, Peptides, and Recombinant Proteins
Lipofectamine RNAiMAX transfection reagent	ThermoFisher Scientific	Cat# 13778150
Lipofectamine 2000 transfection reagent	ThermoFisher Scientific	Cat # 11668019
Fluorophosphonate-biotin	Santa Cruz Biotechnologies	Cat# sc-215056A
ActivX TAMRA-FP Serine Hydrolase Probe	ThermoFisher Scientific	Cat# 88318
Streptavidin-agarose beads	ThermoFisher Scientific	Cat# 20353
WWL113	Sigma-Aldrich	Cat# SML1179
MJN110	Kind gift of Dr. Benjamin Cravatt, Scripps Research Institute (Niphakis et al., 2013)	N/A
Arachidonic acid-d8	Cayman	Cat# 390010-1
2-AG-d5	Cayman	Cat# 362162
Critical Commercial Assays
RNeasy isolation kit	Qiagen	Cat# 74136
10% TGX Stain-Free™ FastCast™ Acrylamide gel	Bio-Rad	Cat# 1610183
Trans-Blot® Turbo™ Transfer System	Bio-Rad	Cat# 1704272
Pierce C18 spin columns	Thermo Fisher	Cat# 89873
iScript cDNA Synthesis Kit	Bio-Rad	Cat# 1708841
SSO advanced SYBR Green Supermix	Bio-Rad	Cat# 1725274
Single Cell RNA purification kit	Norgen Biotek	Cat# 51800
Deposited Data
miR-185 microarray data	(Singaravelu et al., 2015)	NCBI GEO #GSE73165
miR-185 ABPP mass spectrometry data	This paper	Mendeley Data: DOI:10.17632/2h8mf2k3kt.1
Experimental Models: Cell Lines
Huh7.5	Kind gift from Dr. C.M. Rice, Rockefeller University, NY, USA	RRID: CVCL_7927
E9 (Huh7 pFK-I389neo/luc/NS3-3’/5.1)	(Sagan et al. 2006)	N/A
Oligonucleotides
Primers for 18S, see table S4	This paper	N/A
Primers for AADAC, see table S4	This paper	N/A
Primers for ABHD6, see table S4	This paper	N/A
Primers for ACOT1, see table S4	This paper	N/A
Primers for CES1, see table S4	This paper	N/A
Primers for FASN, see table S4	This paper	N/A
Primers for LIPC, see table S4	This paper	N/A
Primers for MGLL, see table S4	This paper	N/A
Primers for PPARα, see table S4	This paper	N/A
Primers for JFH1 IRES, see table S4	This paper	N/A
Recombinant DNA
mirVana hsa-miRNA-185-5p miRNA mimic, assay ID MC12486	ThermoFisher Scientific	Cat# 4464067
mirVana miRNA Mimic, Negative Control	ThermoFisher Scientific	Cat# 4464060
MGLL 3’UTR pEZX-MT06 plasmid	GeneCopoeia	Cat# HmiT001605-MT06
Silencer™ Select Pre-Designed siRNA (ABHD6)	ThermoFisher Scientific	Cat# 4392420 - s33001
Silencer™ Select Pre-Designed siRNA (MGLL)	ThermoFisher Scientific	Cat# 4390824 - s22379
Software and Algorithms
Proteome Discoverer	ThermoFisher Scientific	https://www.thermofisher.com/ca/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/multi-omics-data-analysis/proteome-discoverer-software.html
Transcriptome Analysis Console	Affymetrix, ThermoFisher Scientific	https://www.thermofisher.com/ca/en/home/life-science/microarray-analysis/microarray-analysis-instruments-software-services/microarray-analysis-software/affymetrix-transcriptome-analysis-console-software.html
Graph Pad Prism	GraphPad Prism Software	https://www.graphpad.com/scientific-software/prism/

Significance:

microRNAs are small non-coding RNAs which act as endogenous regulators of gene expression. They simultaneously target multiple messenger RNAs sequence-specifically to prevent translation, thereby affecting downstream cellular pathways and processes. In healthy states, microRNAs are implicated in development, signaling, metabolism and maintaining homeostatic balance. However, they have also been shown to play central roles in disease progression and immunity. Due to their broad effects on multiple targets and signaling cascades, characterizing microRNA function is challenging. Herein, we apply activity-based protein profiling (ABPP) to identify metabolic serine hydrolases whose activity is differentially regulated by microRNA-185, an immunometabolic antiviral microRNA. Using this approach, metabolic enzymatic activity can be probed while accounting for post-translational processing, secondary signaling and indirect effects of microRNA-185. We complement this technique with transcriptomic and lipidomic analysis, linking the proteomic changes to lipid levels, and providing a better understanding of the cellular changes elicited by the microRNA. The study demonstrates the use of ABPP to annotate microRNA function and brings forth new druggable targets for pharmacological recapitulation of non-coding RNA effects. We demonstrate the later through inhibition of the miRNA-185 target monoacylglyceride lipase, showing antiviral effects against the hepatitis C virus. Therefore, this approach allows the identification of major microRNA nodes, which similarly to conventional signaling pathways, can be pharmacologically targeted with compelling therapeutic potential.